COMMENT
Limnol.
Oceanogr., 40(7), 1995, 1336-1343
Society of Limnology
0 1995, by the American
and Oceanography,
Inc.
Status of the iron hypothesis after the Open-Ocean Enrichment Experiment’
John Martin’s vision, intellect, and drive resulted in a
remarkable achievement-an
unprecedented experiment
to test his idea that iron limitation can control the biomass
and productivity
of phytoplankton
in the open ocean
(Martin et al. 1994). The result would surely have gratified
him immensely, and rightly so: after a patch of surface
water in the equatorial Pacific Ocean was enriched with
iron, chlorophyll concentration doubled and primary productivity increased by a factor of four. Clearly, iron had
affected the net growth of phytoplankton.
The experiment (IronEx), described as a test of the iron
hypothesis, represents a milestone in a line of research
(see de Baar 1994) that traces to the founders of biological
oceanography and owes its prominence to the efforts of
Martin and his colleagues (Martin and Fitzwater 1988;
Martin 1992). Enrichment of the open ocean, a contentious proposition, was undertaken only after vigorous debate before and during a special symposium organized
by the American Society of Limnology and Oceanography
(ASLO). The debate was fruitful indeed; a broad crosssection of the oceanographic community reported their
results and discussed their ideas (Chisholm and Morel
199 l), and consensus was reached on a variety of issues,
including the need for a modestly scaled ocean enrichment experiment. It seems clear that open discussion of
the controversial topic, even by those who presented ideas
and alternate interpretations rather than original results,
was healthy. In that spirit, I offer some comments on the
iron hypothesis as research on the topic enters a new stage.
Experiments in bottles failed to resolve questions about
Martin’s iron hypothesis, principally because natural rate
processes, especially grazing, could not be characterized.
In principle, open-ocean iron enrichment would permit
a robust test of the hypothesis: if iron were delivered to
the photic zone in an available form and nothing happened, the hypothesis would be soundly rejected. The
observed response -rapid physiological changes in phytoplankton followed by an increase in their biomass and
productivity - supports Martin’s iron hypothesis, but, as
is discussed below, it leaves many questions unanswered.
The results of IronEx did falsify an alternate, less comprehensive and evocative hypothesis: biomass and productivity of phytoplankton
in the equatorial Pacific are
controlled entirely by factors other than the availability
of iron.
Defining the iron hypotheses
The HNLC-iron
It can be argued that effective research requires formal
and systematic construction of multiple, falsifiable hypotheses (Platt 1964) and that precision is essential when
formulating or discussing hypotheses about the control
of primary production (Cullen 199 1). Likewise, hypotheses can be clarified by careful review of their historical
context (de Baar 1994), and they can be refined or questioned by exploring the processes responsible for experimental results (Banse 199 1). Another approach is to pursue a powerful idea with unwavering purpose, developing
new techniques, collecting novel data, and compiling a
broad range of supporting evidence from other sources.
Martin and his colleagues used the latter approach effectively, making a strong case for iron limitation
in the
major nutrient-rich waters of the sea (Martin et al. 199 1).
In most discussions of the iron hypothesis, the principal
unexplained phenomenon is the existence of high-nutrient, low-chlorophyll
(HNLC) waters in the open ocean
(Chisholm and Morel 199 1). Accordingly, at an early stage
in the discussion of Martin’s ideas (October 1990), participants at a workshop convened by the Board on Biology
of the U.S. National Research Council formulated what
I will call the HNLC-iron hypothesis: “An increase in the
rate of supply of iron to the surface layer of the ocean
will reduce to depletion the unused macronutrients,
nitrate and phosphate” (see Cullen 199 1, p. 1585). It was
intended to be a testable explanation of the HNLC condition. During IronEx, the supply of iron to the surface
layer was increased, but the concentrations of nitrate remained high and changes of CO2 were minimal (Watson
et al. 1994). At face value, the HNLC-iron
hypothesis
was rejected; however, neither the hypothesis nor the test
were perfect. The hypothetical increase in the rate of sup-
1 Accepted
13 April
1995.
1336
The iron hypothesis is not explicitly defined by Martin
et al. (1994), but the experiment clearly was testing the
proposition “that phytoplankton
growth in major nutrient-rich waters is limited by iron deficiency” (Martin et
al. 199 1, p. 1793). Let us define this as Martin’s iron
hypothesis and broaden the discussion to include other,
more specific hypotheses, as well as alternate hypotheses.
Recent reviews (de Baar 1994; Geider and La Roche 1994)
can be consulted for background, insight, and more comprehensive descriptions of relevant research.
Martin’s iron hypothesis
hypothesis
Comment
ply of iron is not defined, and regardless, the open-ocean
addition of iron was too ephemeral to constitute an unequivocal test of the HNLC-iron
hypothesis.
Now that the first results of IronEx have been reported,
the HNLC-iron hypothesis is in a state of tantalizing uncertainty. The fact is that the concentrations of nitrate
and phosphate changed little in response to the iron addition. Explanations for the phenomenon differ (Kolber
et al. 1994; Martin et al. 1994; Van Scoy and Coale 1994;
Watson et al. 1994; Banse 1995). These are discussed
below in the context of alternate working hypotheses.
The ecumenical iron hypothesis
By 199 1, when the issue was finally discussed in a broad
public forum (Chisholm and Morel 199 l), those who were
inclined to explore the ecological bases of Martin’s iron
hypothesis had developed remarkably consistent explanations for the HNLC condition. Morel et al. (199 lb)
called it the ecumenical iron hypothesis. Different proponents (e.g. Banse 1990, 199 1; Chavez et al. 199 1; Cullen 199 1; Frost 199 1; Miller et al. 199 1; Price et al. 199 1;
Cullen et al. 1992a; DiTullio
et al. 1993) emphasized
particular aspects of the hypothesis, but in general they
suggested that when iron is scarce, the dominant smaller
cells with greater surface:volume ratios can grow more
rapidly than larger cells (Morel et al. 199 la; Chisholm
1992). The specific growth rates of small cells are not
strongly limited by iron (Price et al. 199 1, 1994; Cullen
et al. 1992a); rather, their numbers are controlled by
microzooplankton
grazers whose potentially high growth
rates (Banse 1982) enable them to keep small phytoplankton populations in check (Frost 199 1; Miller et al. 199 1).
Larger cells cannot attain high growth rates at ambient
nutrient concentrations, but enrichment with iron would
allow them to grow and assimilate nitrate, unfettered by
microzooplankton
grazing because of their large size, and
unchecked by mesozooplankton
grazing because those
herbivore populations could not respond in time. Clearly,
this latter supposition cannot be tested with incubation
experiments.
The ecumenical iron hypothesis views the HNLC condition as “grazer controlled phytoplankton
populations
in an iron-limited
ecosystem” (Price et al. 1994, p. 520).
Although this hypothesis provides an excellent context
for experimentation
and discussion (Barber and Chavez
199 1; Chavez et al. 199 1; Price et al. 199 1, 1994; Cullen
et al. 1992a), it is not amenable to simple comprehensive
tests and does not explain all results from enrichment
experiments (see Coale 199 1). The ecumenical iron hypothesis suggests that no single factor regulates primary
productivity;
rather, the interplay of factors is key (see de
Baar et al. 1990). A goal, then, is to see whether iron plays
a disproportionate
role in determining the use of macronutrients (Geider and La Roche 1994).
The ecumenical iron hypothesis does not require that
small cells grow at nutrient-saturated
specific growth rates
(II.max9 * d-l). Rathe r, it states that the specific growth rates
(CL)of small cells are not strongly limited by iron, hence
1337
grazing loss (g; d-l) is the predominant control on their
net rate of increase (k; d-l). This criterion is not very
precise, so the distinction between physiological limitation by iron and grazing control of small cells is fuzzy at
best. Consider HNLC waters where k is normally near
zero, and grazing is the dominant loss term for small cells
(see Frost 199 1; Cullen 199 1): k M p - g, thus h M g.
We can define the relative influence of physiological limitation on k as (1 - ~I~,,,), and the relative influence of
grazing as (g/pm,,). One would thus have to demonstrate
strong iron limitation of p (i.e. p < 0.5 p,.,.,,,)to infer that
iron is the principal mechanism controlling net rates of
growth. In HNLC waters of the equatorial Pacific, photochemical energy conversion efficiency is about half
maximal and it increases during incubations with added
iron (Greene et al. 1994). These observations are interpreted by the investigators as evidence of physiological
control of photosynthesis and growth of phytoplankton.
However, all,,, in situ is uncertain, and it is known that
phytoplankton
under “metal stress” can have strongly
altered physiological characteristics even though growth
rates exceed 90% of maximum (Morel et al. 199 1a). The
study of Greene et al. (1994) leaves us with a strong
indication that specific growth rates of equatorial phytoplankton are less than maximal, but no accurate knowland no direct information
on how much
edge
of dpmax
k and g would change if iron-limited specific growth rates
of small cells increased in response to iron enrichment.
Their study is an excellent segue to the open-ocean enrichment experiment.
One prominent feature of IronEx - an apparently strong
physiological response to iron by small phytoplankton
with significant and similar net increases of chlorophyll
in small as well as larger size classes (Kolber et al. 1994)is consistent with the suggestion that the photosynthesis
and biomass of small cells is controlled by physiological
limitation (Greene et al. 1994). This view is contrary to
the central tenet of the ecumenical iron hypothesis, that
larger cells will dominate the response to iron enrichment.
Cell counts and biomass estimates for all size classes (including prochlorophytes)
should help resolve the degree
to which iron enrichment influenced the biomass, as compared to the chlorophyll content of each size class.
As results are presented and discussed in future reports,
it will be natural to compare results from IronEx to those
from elsewhere in the equatorial Pacific. In doing so, it
should be remembered that micronutrient
inputs to the
equatorial Pacific might be episodic and patchy (Donaghay et al. 199 1) and, consequently, the nutritional status
of natural phytoplankton
assemblages may vary in space
and time. We should thus recognize that the results of
bottle experiments, IronEx, or subsequent enrichment experiments might not apply generally to the HNLC equatorial Pacific.
The grazing hypothesis
Although one could construct a grazing hypothesis that
is exclusive of Martin’s iron hypothesis, it is well recognized that grazing is but one factor controlling the bio-
Comment
1338
Primary Productivity (mg C mm3de’)
0
10
20
30
40
result of IronEx was validation of the grazing hypothesis!
Of course, it’s more complicated than that.
50
Results of IronEx compared to natural variability
-
50
60
out
+300
In
-301
In
85"-9O"W
1.111111.
g(pg5ow
Fig. 1. Primary productivity at and near the site of the openocean enrichment experiment (near 5’S, 9OOW). Profiles from
out of the patch and in the patch 2 d (calendar day 300) and
3d (calendar day 301) after enrichment are from Martin et al.
(1994). Profiles of historical averages east (4-6’S, 85-9O”W; n
= 10) and west of the site (4-6”S, 90-95”W; n = 11) are from
R. Barber and F. Chavez as presented by Martin and Chisholm
(1992). Error bars for the measurements during IronEx were
presented by Martin et al. (1994) but not defined. For the average
profiles, errors (presumed to be SE) were 16-22% (X = 18%) of
the mean for 85-9O”W and 7-22% (X = 13%) for 90-95”W.
mass and productivity of phytoplankton (Walsh 1976; de
Baar et al. 1990; Frost 199 1). Nonetheless, there is good
reason to conclude that in HNLC waters grazing keeps
phytoplankton standing crops at a lower level than would
be attained had available iron been completely consumed.
Specific growth rates of phytoplankton are relatively high
(Banse 199 1; Miller et al. 199 1; Cullen et al. 1992a),
standing crops are fairly constant (Frost 199 l), and during
incubation experiments (in which natural grazing is disrupted), unenriched phytoplankton
in the controls grow
to higher concentrations than are observed in the natural
systems (Buma et al. 199 1; Price et al. 199 1; de Baar
1994). Why are biomass levels poised at those concentrations? With respect to iron hypotheses, we can ask, “If
iron supplies are enhanced, and the specific growth rates
of phytoplankton
increase, will standing stocks increase
to the extent that macronutrients are consumed, or will
grazers keep phytoplankton
biomass low?” For an openocean enrichment experiment, an appropriate grazing hypothesis would be that after enrichment with iron, the
reiponse of grazers would prevent the biomass of phytoplankton from increasing enough to deplete macronutrients. Given the evidence of significant increases in grazing pressure (discussed below), along with the minimal
depletion of nutrients despite stimulated growth of phytoplankton, it would seem at face value that a principal
The response of equatorial Pacific surface waters to
added iron was evaluated principally by comparison to
waters outside an enriched patch (Kolber et al. 1994;
Martin et al. 1994; Watson et al. 1994). Measurements
upstream and downstream of the Galapagos Islands provided an important adjunct: they described changes associated with a natural perturbation
that included enrichment of surface waters with iron.
One could also evaluate the results of IronEx by relating
the changes associated with iron enrichment to natural
variability
in the study region. For a first look, we can
examine background data used in the planning of the
enrichment experiment (Martin and Chisholm 1992).
These data suggest that primary productivity
east of the
IronEx site is generally higher than to the west (Fig. 1)
and that there can be substantial spatial variability
in
primary productivity.
Temporal variability
associated
with El Niiio Southern Oscillation is also expected (Barber
et al. 1994).
It seems that the open-ocean enrichment was performed on a particularly impoverished parcel of water:
near-surface productivity outside the iron-enriched patch
was much lower than the historical averages (Fig. 1).
Anomalously low productivity
may be a clue as to why
small phytoplankton
seemed to be iron limited at the
IronEx site (Kolber et al. 1994) but not during other studies in the equatorial Pacific (Price et al. 199 1,1994; Cullen
et al. 1992a). When iron limitation was relieved by enrichment, productivity
was elevated close to historical
averages for the region (Fig. 1). Failure to achieve rates
substantially higher than what occurs naturally makes it
difficult to reject the hypothesis that something other than
iron has a strong influence on rates of productivity
in the
region.
Patterns of PB [productivity
normalized to chlorophyll;
g C (g Chl)- l d- ‘1 prior to and during IronEx (Fig. 2) merit
examination. Interpreting vertical patterns is complicated
because near-surface rates can be severely underestimated
when samples from mixed layers are incubated all day at
surface irradiance (Marra 1978; Cullen and Lewis 1995),
as was done during IronEx. The maximum PBin the water
column thus may be more appropriate than surface-layer
averages for comparison between sites and times (cf. Smith
et al. 1980; Murray et al. 1994). Maximum values suggest
that during IronEx, PB started out unusually low and was
elevated by iron enrichment to the historical average for
the region. The pattern is consistent with that for productivity but is not easily interpreted in terms of phytoplankton growth. Remember that specific growth rate
(p.;d-‘) = PB/C:Chl,
w h ere C : Chl is the ratio of cellular
carbon to chlorophyll [g C (g Chl))‘]. As pointed out by
Martin et al. (1994), specific growth rates of phytoplankton accelerated in response to iron, but PB did not change
as much, because the ratio of C: Chl decreased. Such
Comment
nutrition-related
changes in C : Chl can completely eliminate any relationship between PB and p for phytoplankton in balanced growth (Sakshaug et al. 1989). Productivity normalized to chlorophyll is thus inappropriate as
a general diagnostic of nutrient limitation
(Cullen et al.
1992b; Falkowski et al. 1992).
IronEx showed that PB was relatively insensitive to
changes in iron nutrition
because there were complementary changes of specific growth rate and C : Chl (see
also Coale 199 1). Changes of PBindependent of p indicate
that previous analyses of equatorial Pacific growth processes, in which PBwas considered proportional to growth
rate (Barber and Chavez 199 1; Murray et al. 1994), should
be reconsidered. Short-term changes in PB are reflected
to some extent in growth rates, but relatively constant PB
in a region does not mean that growth rates are likewise
constant (Cullen et al. 1992b). Thus, the relative uniformity of water-column PB in the eastern equatorial Pacific
Ocean (Barber and Chavez 199 1) may mask considerable
variability
in the nutrient-limited
growth rates of phytoplankton. It follows that future enrichment experiments
may encounter fundamentally different populations than
were enriched during IronEx; if responses of assemblages
differ between experiments, it will be a challenge to determine the extent to which initial conditions vs. experimental differences influenced the results.
Interpreting IronEx in the context of iron hypotheses
The principal result of IronEx was exciting and unequivocal: a patch of water was enriched with iron and
the phytoplankton responded with an increase in biomass
and productivity. However, the bloom terminated quickly, macronutrients
were hardly depleted, and CO, was
only minimally drawn down (Watson et al. 1994). Why?
Martin et al. (1994) presented a thoughtful discussion
with reference to paralkl studies in the Galapagos plume.
Additional
information
and interpretations
have been
published by IronEx participants (Kolber et al. 1994; Van
Scoy and Coale 1994; Watson et al. 1994), and Banse
(1995) has elaborated the importance of zooplankton.
Here, I offer another perspective, based only on information published to date.
As discussed by Martin et al. (1994), at least three
hypotheses could explain why the response of phytoplankton to iron enrichment did not culminate in a larger
bloom and depletion of macronutrients:
another micronutrient ran short; iron was lost to the system; and grazing
increased rapidly. Let us consider these explanations.
Depletion of a second micronutrient -Martin
et al.
(1994) discounted, but did not entirely exclude the possibility that a second micronutrient
was depleted in the
enriched patch. They argued that there were no significant
depletions in dissolved Zn, Cu, Ni, Co, or Cd within the
patch and that macronutrients
are depleted (i.e. other
trace elements do not become limiting) in trace-metal
clean bottle incubations with additions of iron alone (e.g.
Martin et al. 199 1; Price et al. 1994). These are good
arguments for an early stage of analysis. If possible, it
1339
PB (g C g Chl-’ d-‘)
0
40
50
60
20
40
60
80
-out
+300
In
+301
In
-85'9O"W
1.1.11.1.90" _ g5ow
Fig. 2. Primary productivity
normalized to chlorophyll a at
and near the site of the open-ocean enrichment experiment.
Locations, symbols, and sources as in Fig. 1. In the tables presented by Martin and Chisholm (1992), the depths and stations
for average chlorophyll concentrations were not the same as for
productivity, so estimates of PB for 85-9O”W and 90-95”W were
obtained using depth-interpolated
values of chlorophyll.
would be useful to determine the complexation of trace
elements (Bruland et al. 199 1) in and out of the patch.
Also, as discussed above, it would be prudent to consider
that results from experiments elsewhere in the equatorial
Pacific might not apply to IronEx, and vice versa. For
example, Wells et al. (1994) found contrasting responses
to iron in two different assemblages from the equatorial
Pacific. They interpreted one response as being consistent
with depletion of a micronutrient other than iron. It seems
reasonable that iron nutrition, and perhaps ratios of traceelement bioavailability,
are variable in HNLC waters affected by aeolian input and variable influence of the equatorial undercurrent (see Martin 1992; Murray et al. 1994).
At this time, we know very little about the patchiness of
bioavailable micronutrients
in HNLC waters (Donaghay
et al. 199 1). It is clear, however, that trace-element interactions can be quite complicated in the open ocean
(Bruland et al. 199 1).
Loss of iron from the system-A second explanation,
much more consistent with the ideas of Martin and his
colleagues, is that “iron was lost from the system due to
colloidal aggregation and/or sinking of large particles containing iron” (Martin et al. 1994, p. 128). Martin et al.
(1994) did not elaborate on how colloidal aggregation may
have led to loss of iron from the system. The transformations and availability of iron are being actively studied
(e.g. Hutchins et al. 1993; Johnson et al. 1994; Wells et
al. 1994) though, and the complex interactions between
competing processes will surely be addressed in the future.
1340
Comment
Sinking of large particles was carefully considered by
Martin et al. (1994). They pointed out that the sinking
of large phytoplankton,
specifically diatoms, can keep the
population low, thereby limiting absolute rates of nutrient
uptake by preventing the accumulation of photosynthetic
biomass in the photic zone. This hypothesis, related to
ideas presented by Chavez et al. (199 I), bears careful
examination because it may be key to the eventual explanation of ecosystem response to iron.
First, we should ask whether sinking of diatoms could
be responsible for the rapid loss of iron from the systema loss that may have prevented the complete utilization
of macronutrients.
Remember that the iron enrichment
was more than adequate to support complete utilization
of macronutrients if the system had responded like a bottle experiment (initial nitrate = 10.8 PM, Fe on day 2 =
3.6 nM: Martin et al. 1994; N : Fe of 5,000: Martin 1992).
If diatoms took up most of the iron and sank from the
system, they would have brought stoichiometric
equivalents of macronutrients and carbon dioxide with them.
However, only about 7 PM CO2 was utilized (Watson et
al. 1994), hence 1 PM nitrate, so the N : Fe ratio of the
sinking particles would have been 280, or a factor of 17
lower than expected. Perhaps large amounts of iron were
adsorbed to sinking diatoms and other particles, such as
fecal pellets. Regardless, it seems likely that other processes besides biological uptake and sinking were responsible for the rapid loss of iron from the IronEx system.
The sinking of diatoms may still be important in explaining the feeble utilization
of macronutrients
during
IronEx. As pointed out by Martin et al. (1994), if sinking
had kept the population of diatoms low, then absolute
rates of macronutrient uptake would be less than if diatoms had accumulated. If the only losses of iron to the
system had been associated with the sinking of biogenic
particles (i.e. C, N, and P are exported with the sinking
Fe), then at first order, sinking would influence the rate
of macronutrient depletion, but not the amount of macronutrients consumed per unit of added iron. However,
if iron were simultaneously being removed from the system in a competing process without, concomitant removal
of macronutrients, then sinking would definitely influence
the stoichiometry between iron additions and the utilization of macronutrients and CO,. Much greater amounts
of iron would be necessary to satisfy the HNLC-iron
hypothesis, and processes that influence the net growth of
phytoplankton assemblages (e.g. grazing, sinking, and light
limitation associated with the observed subduction of the
Fe-enriched patch) would have to be considered quantitatively in a robust analysis of results. It will be interesting to see how observations of large losses of iron from
the IronEx system are reconciled with results indicating
rapid and efficient recycling of iron in equatorial Pacific
HNLC waters (Hutchins et al. 1993).
Grazing control-The
HNLC-iron
hypothesis and the
ecumenical iron hypothesis would be falsified if grazers
checked the net growth of phytoplankton,
thereby preventing nutrient depletion in iron-enriched
HNLC wa-
ters. Watson et al. (1994, p. 145), using the parlance of
geochemists, suggested that this is exactly what happened
during IronEx: “after a brief [period] of disequilibrium,
the ecosystem responded by recycling carbon rapidly back
to the inorganic form.” Calculations of grazing losses
(Banse 1995) also support the hypothesis of grazing control. Although grazing was not quantified directly, there
were indications that it increased in the patch. Not only
did microheterotrophic
biomass jump 50%, mesozooplankton biomass in the surface layer increased and vertically migrating zooplankton apparently remained in the
patch during the day rather than descending as they normally do (Martin et al. 1994; Van Scoy and Coale 1994;
see also Banse 1995).
As recognized by Donaghay et al. (199 l), rapid responses of zooplankton to episodic nutrient enrichments
may keep levels of phytoplankton biomass relatively low.
Martin et al. (1994) accepted that increased grazing exerted some control on the IronEx bloom, but argued that
grazing was not a likely mechanism for preventing depletion of major nutrients. Rather, retention of iron in
the system was seen as the critical factor. Shallow waters
near the Galapagos Islands were presented as an example
of natural waters in which nitrate was depleted because
iron was supplied continuously.
The key assumption is
that the potential for enhanced grazing is similar over the
Galapagos platform as compared to the IronEx site. However, the relatively restricted, shallow and hydrographitally complex region where nitrate depletion was observed (see Chavez and Brusca 199 1) may not fully represent the open-ocean ecosystem that was experimentally
enriched. For example, it is not clear that the increases
of mesozooplankton
biomass within the iron-enriched
patch, apparently associated with altered vertical migration (anticipated by Donaghay et al. 199 1 and reported
by Van Scoy and Coale 1994), were or could be a feature
of the nitrate-depleted, iron-rich waters of the Galapagos
platform. No doubt, careful assessment of grazing will
figure in future studies of iron enrichment. Grazing pressure can be measured by various means (see Murray et
al. 1994), but it might be difficult to interpret the responses of zooplankton to a localized enrichment because
the purported enhancement of grazing pressure by altered
migration patterns should be a scale-dependent phenomenon.
Independent arguments have been presented to discount the role of grazing in limiting the IronEx bloom.
Kolber et al. (1994) assessed the photosynthetic
physiology of phytoplankton using sensitive fluorescence methods (Kolber and Falkowski 1993). The response within
the patch was rapid, striking, and strongly consistent with
expectations based on laboratory studies of phytoplankton relieved from iron starvation (Greene et al. 1992).
The evidence for higher specific growth rates in the patch
is convincing. The observed increase of biomass was not
sustained for long, however, Kolber et al. (1994) noted
that net growth may have halted not because of grazing,
but because the specific growth rates of phytoplankton
slowed, either due to limitation by another trace element
Comment
or because iron ran out. Measurements of photosynthetic
energy conversion efficiency (FJAF,; maximum value in
the upper 30 m of the water column) increased sharply
after iron enrichment, then declined after day 2, with a
reported e-folding time of 6 d. However, during this same
period, the iron-enriched patch was subducted to a depth
of 30-35 m (Martin et al. 1994), so observations in the
upper 30 m had shifting relevance. Also, two other measures related to iron limitation (rQA and aps2; see Kolber
et al. 1994) showed little or no reversion to the pre-enriched state after day 2. More extensive analysis will be
required to make a convincing refutation of grazing control.
The iron hypothesis reconsidered
The grazing issue may be clarified soon. For now, it
seems that the insignificant consumption of macronutrients during IronEx, apparently associated with increased
grazing and subduction of the Fe-enriched patch, is a
strong indication that even though iron can stimulate
growth of HNLC phytoplankton,
grazing may prevent
depletion of macronutrients or at least greatly retard the
utilization of major nutrients during the long residence
of HNLC waters near the surface (Minas et al. 1986;
Fiedler et al. 199 1). Such effective control of nutrient
utilization by grazing, if demonstrated rigorously, would
lead to rejection of the HNLC-iron
hypothesis, but it
would leave Martin’s more general iron hypothesis intact.
Perhaps it is time to recast the iron hypothesis. The
shift away from HNLC conditions toward the blooming
of phytoplankton as the phenomenon to be explained by
the iron hypothesis is reflected in a recent study by de
Baar et al. (1995, p. 4 12), who framed the iron hypothesis
as a “suggestion that iron is a limiting nutrient for plankton productivity
and consequent CO2 drawdown.” They
explored the role of iron in the Southern Ocean-another
important HNLC region - where several factors, including light, are thought to influence primary production
(Dugdale and Wilkerson 1990; Buma et al. 199 1; Mitchell
et al. 199 1; Nelson and Smith 199 1). By relating natural
levels of chlorophyll, primary productivity,
macronutrients, and CO, to distributions of iron in open-ocean Antarctic waters, de Baar et al. (1995) made a case for the
regulation of diatom blooms by iron. They asserted that
blooms can occur in the iron-rich jet of the polar front,
but not in the iron-poor waters of the Antarctic circumpolar current. Their study of the polar jet and adjacent
waters can be compared to the work in and near the
Galapagos reported by the IronEx team. A difference is
that the Antarctic waters are truly oceanic.
de Baar et al. (1995) made no claims that iron is solely
responsible for high nutrient conditions in the Southern
Ocean. Rather, they argued that iron is biogeochemically
important because it permits significant drawdown of surface CO2 in blooms. The blooms they discuss (Chl a
concentrations of 2-3 mg m-3) consumed only a small
fraction of available nitrate, phosphate, and silicate. As
1341
de Baar et al. (1995) acknowledged, other factors, such
as vertical mixing and grazing, prevent complete utilization of nutrients (see also de Baar et al. 1990; Buma et
al. 1991). A corollary is that iron is not predominately
responsible for the HNLC condition in those waters (see
Mitchell et al. 199 1; Nelson and Smith 1991).
Regardless of unresolved issues, the results of IronEx
and the study by de Baar et al. (1995) signal an important
shift in the iron hypothesis. The HNLC-iron
hypothesis
has not been strongly supported nor has it been conclusively rejected. It seems that future studies will focus on
the degree to which surface concentrations of macronutrients and CO2 are depleted as a function of iron supply,
without the need to demonstrate complete depletion of
nutrients. Controls on the rate of macronutrient depletion
(a community-level
process; Banse 1995) and on the losses
of iron from surface waters, as well as the residence time
of surface waters (Minas et al. 1986), must be considered
in order to explain the role of iron in HNLC waters. The
work will be difficult and complicated but exciting, involving the collaboration of “clever biologists and skillful
chemists” (Bruland et al. 199 1, p. 1574). Open discussion
and international, interdisciplinary
collaboration will help
in planning, implementation
and interpretation of results.
Future directions
The role of iron in ocean biology has been studied for
many years, but the most rapid progress, made possible
by technological advance has occurred during the past
decade, propelled by the efforts of Martin and his colleagues and strengthened by contributions
from many
other researchers. As judged by the course of events, the
open discussion and vigorous debate at the ASLO special
symposium facilitated this progress and paved the way
toward IronEx, even though the experiment stayed true
to the ideas of those who originally proposed the work
(Watson et al. 199 1; Martin 1992). There is clear justification for more open-ocean enrichment experiments.
Likewise, there are good reasons to pursue thorough studies of natural enrichment, such as in the Galapagos plume
and the Antarctic
polar jet. Interpretations
can be
strengthened through the expression of numerous points
of view.
It will be healthy for a broad spectrum of marine scientists to become involved in discussing published results
and suggesting designs for future experiments. Here is a
suggestion. Moor several barges in open, high-nutrient
waters of the equatorial Pacific. Attach several thousand
kilograms of metallic iron on each, engineered with plenty
of surface area to encourage corrosion (rusting) at a predetermined, environmentally
relevant rate. Incorporate
an inert tracer, if possible (Watson et al. 199 1). Treat the
persistent enrichment site as an international
scientific
resource, and encourage as many groups as possible to
study the chemical and ecological changes associated with
the iron plume. Such an open research plan would encourage a broad diversity of approaches and interpreta-
1342
Comment
tions, leading, one would hope, to rapid progress toward
resolving a key issue in modern oceanography. The iron
hypothesis might change in the process, but its central
message, promoted so well by John Martin, is likely to
survive.
John J. Cullen
Department of Oceanography
Dalhousie University
Halifax, Nova Scotia B3H 451
References
BANSE, IS. 1982. Cell volumes, maximal growth rates of unicellular algae and ciliates, and the role of ciliates in the
marine pelagial. Limnol. Oceanogr. 27: 105 9- 107 1.
. 1990. Does iron really limit phytoplankton production
in the offshore subarctic Pacific? Limnol. Oceanogr. 35:
772-775.
-.
199 1. Rates of phytoplankton
cell division in the field
and in iron enrichment experiments. Limnol. Oceanogr.
36: 1886-l 898.
1995. Community response to IRONEX. Nature 375:
112.
BARBER, R. T., AND F. P. CHAVEZ. 199 1. Regulation of primary
productivity
rate in the equatorial Pacific. Limnol. Oceanogr. 36: 1803-1815.
-,
J. W. MURRAY, JR., AND J. J. MCCARTHY. 1994. Biogeochemical interactions in the equatorial Pacific. Ambio
23: 62-66.
BRULAND, K. W., J. R. DONAT, AND D. A. HUTCHINS. 1991.
Interactive influences of bioactive trace metals on biological
production in oceanic waters. Limnol. Oceanogr. 36: 15 5 51577.
BUMA, A. G. J., H. J. W. DE BAAR, R. F. NOLTING, AND A. J.
VAN BENNEKOM. 199 1. Metal enrichment experiments in
the Weddell-Scotia Seas: Effects of iron and manganese on
various plankton communities.
Limnol. Oceanogr. 36:
1865-l 878.
CHAVEZ, F. P., AND R. C. BRUSCA. 199 1. The Galapagos Islands and their relation to oceanographic processes in the
tropical Pacific, p. 9-33. In M. J. James [ed.], Galapagos
marine invertebrates. Plenum.
, AND OTHERS. 199 1. Growth rates, grazing, sinking, and
2 Joint affiliation: Bigelow Laboratory
West Boothbay Harbor, Maine 045 75.
for Ocean Sciences,
Acknowledgments
I thank the Office of Naval Research, NASA, and NSERC for
support; K. Banse, R. F. Davis, H. de Baar, G. A. Jackson, M.
E. Webb, and an anonymous reviewer for comments; S. W.
Chisholm for a review; and D. Kirchman for editorial comments
and the invitation to present these ideas. K. H. Coale’s review
contained many useful comments and suggestions, which are
much appreciated. However, some differences of perspective or
opinion remain. Certain ideas presented by reviewers and colleagues would have improved this commentary, but the ideas
were based on data not yet published and are not presented here.
iron limitation of equatorial Pacific phytoplankton.
Limnol. Oceanogr. 36: 1816-1833.
CHISHOLM, S. W. 1992. Phytoplankton
size, p. 213-238. In
P. G. Falkowski and A. Woodhead [eds.], Primary productivity and biogeochemical cycles in the sea. Plenum.
AND F. M. M. MOREL [eds.]. 199 1. What controls
phytoplankton production in nutrient-rich areas of the open
sea? Limnol. Oceanogr. 36: 1507-l 965.
COALE, K. H. 1991. Effects of iron, manganese, copper, and
zinc enrichments on productivity
and biomass in the subarctic Pacific. Limnol. Oceanogr. 36: 185 l-l 864.
CULLEN, J. J. 199 1. Hypotheses to explain high-nutrient conditions in the open sea. Limnol. Oceanogr. 36: 1578-l 599.
AND M. R. LEWIS. 1995. Biological processes and optical measurements near the sea-surface: Some issues relevant to remote sensing. J. Geophys. Res. 100: 13,25513,266.
C. 0. DAVIS, AND R. T. BARBER. 1992~. Photosynthetic characteristics and estimated growth rates indicate grazing is the proximate control of primary production in the equatorial Pacific. J. Geophys. Res. 97: 639654.
-,
X. YANG, AND H. L. MACINTYRE. 1992b. Nutrient
limitation
of marine photosynthesis, p. 69-88. In P. G.
Falkowski and A. Woodhead [eds.], Primary productivity
and biogeochemical cycles in the sea. Plenum.
DE BAAR, H. J. W. 1994. von Liebig’s law of the minimum
and plankton ecology (1899-199 1). Prog. Oceanogr. 33:
347-386.
AND OTHERS. 1990. On iron limitation of the Southern
Ocean: Experimental observations in the Weddell and Scotia Seas. Mar. Ecol. Prog. Ser. 65: 105-122.
-,
AND OTHERS. 1995. Importance of iron for plankton
blooms and carbon dioxide drawdown in the Southern
Ocean. Nature 373: 4 12-4 15.
DITULLIO,G. R.,D.A. HUTCHINS,AND K.W. BRULAND. 1993.
Interaction of iron and major nutrients controls phytoplankton growth and species composition in the tropical
north Pacific Ocean. Limnol. Oceanogr. 38: 495-508.
DONAGHAY, P. L., AND OTHERS. 199 1. The role of episodic
atmospheric nutrient inputs in the chemical and biological
dynamics of oceanic ecosystems. Oceanography 4: 62-70.
DUGDALE, R. C., AND F. P. WILKERSON. 1990. Iron addition
experiments in the Antarctic: A reanalysis. Global Biogeothem. Cycles 4: 13- 19.
FALKOWSKI, P. G., R. M. GREENE, AND R. J. GEIDER. 1992.
Physiological limitations on phytoplankton productivity
in
the ocean. Oceanography 5: 84-9 1.
FIEDLER, P. C., V. PHILBRICK, AND F. P. CHAVEZ. 199 1. Oceanic upwelling and productivity
in the eastern tropical Pacific. Limnol. Oceanogr. 36: 1834-1850.
FROST, B. 199 1. The role of grazing in nutrient-rich areas of
the open sea. Limnol. Oceanogr. 36: 16 16-1630.
GEIDER, R. J., AND J: LA ROCHE. 1994. The role of iron in
phytoplankton
photosynthesis, and the potential for ironlimitation of primary productivity
in the sea. Photosynth.
Res. 39: 275-301.
GREENE, R.M.,R.J. GEIDER,Z.KOLBER,AND P.G. FALKOWSKI.
1992. Iron-induced changes in light harvesting and photochemical energy conversion processes in eukaryotic marine algae. Plant Physiol. 100: 565-575.
Z. S.KOLBER, D.G.SWIFT,N. W. TINDALE, AND P.G.
FAKOWSKI.
1994. Physiological
limitation
of phytoplankton photosynthesis in the eastern equatorial Pacific
determined from variability in the quantum yield of fluorescence. Limnol. Oceanogr. 39: 106 l-l 074.
Comment
HUTCHINS, D. A., G. R. DITULLIO, AND K. W. BRULAND. 1993.
Iron and regenerated production: Evidence for biological
iron recycling in two marine environments. Limnol. Oceanogr. 38: 1242-1255.
JOHNSON, K.S.,K.H.
COALE, V.A. ELROD,ANDN. W. TINDALE.
1994. Iron photochemistry in seawater from the equatorial
Pacific. Mar. Chem. 46: 3 19-334.
KOLBER, Z. S., AND OTHERS. 1994. Iron limitation of phytoplankton photosynthesis in the equatorial Pacific Ocean.
Nature 371: 145-149.
-,
AND P. G. FALKOWSKI. 1993. Use of active fluorescence to estimate phytoplankton
photosynthesis in situ.
Limnol. Oceanogr. 38: 1646- 166 5.
MARRA, J. 1978. Effect of short-term variation in light intensity on photosynthesis of a marine phytoplankter:
A laboratory simulation study. Mar. Biol. 46: 19 l-202.
MARTIN, J. 1992. Iron as a limiting factor in oceanic productivity, p. 123-138. In P. G. Falkowski and A. Woodhead
[eds.], Primary productivity
and biogeochemical cycles in
the sea. Plenum.
AND P. CHISHOLM. 1992. Design for a mesoscale iron
emichment experiment. U.S. JGOFS Planning Office Rep.
15. 28 p.
-,
K. H.’ COALE, AND OTHERS. 1994. Testing the iron
hypothesis in ecosystems of the equatorial Pacific Ocean.
Nature 371: 123-l 29.
-,
AND S. E. FITZWATER. 1988. Iron deficiency limits
phytoplankton
growth in the north-east Pacific Subarctic.
Nature 331: 34 l-343.
-,
R.M. GORDON,AND S.E. FITZWATER. 1991. Thecase
for iron. Limnol. Oceanogr. 36: 1793-1802.
MILLER, C. B., AND OTHERS. 199 1. Ecological dynamics in the
subarctic Pacific, a possibly iron-limited
ecosystem. Limnol. Oceanogr. 36: 1600- 16 15.
MINAS, H. J., M. MINAS, AND T. T. PACKARD. 1986. Productivity in upwelling areas deduced from hydrographic and
chemical fields. Limnol. Oceanogr. 31: 1182-1206.
MITCHELL, B.G.,E.A.
BRODY,O.HOLM-HANSEN,~.
MCCLAIN,
AND J. BISHOP. 1991. Light limitation of phytoplankton
biomass and macronutrient
utilization
in the Southern
Ocean. Limnol. Oceanogr. 36: 1662-1677.
MOREL, F. M. M., R. J. M. HUDSON, AND N. M. PRICE. 199 la.
1343
Limitation of productivity
by trace metals in the sea. Limnol. Oceanogr. 36: 1742-l 7 5 5.
-,
J. G. RUETER, AND N. M. PRICE. 199 1b. Iron nutrition
of phytoplankton and its possible importance in the ecology
of ocean regions with high nutrient and low biomass.
Oceanography 4: 5 6-6 1.
MURRAY, J. W., R. T. BARBER, M. R. ROMAN, M. P. BACON,
AND R. A. FEELEY. 1994. Physical and biological controls
on carbon cycling in the equatorial Pacific. Science 266: 5 865.
NELSON, D. M., AND W. 0. SMITH, JR. 199 1. Sverdrup revisited: Critical depths, maximum chlorophyll levels, and the
control of Southern Ocean productivity
by the irradiancemixing regime. Limnol. Oceanogr. 36: 16 50- 166 1.
PLATT, J. R. 1964. Strong inference. Science 146: 347-353.
PRICE, N. M., B. A. AHNER, AND F. M. M. MOREL. 1994. The
equatorial Pacific: Grazer controlled phytoplankton
populations in an iron-limited
ecosystem. Limnol. Oceanogr.
39: 520-534.
-,
L. F. ANDERSEN, AND F. M. M. MOREL. 1991. Iron
and nitrogen nutrition of equatorial Pacific plankton. DeepSea Res. 38: 136 l-l 378.
SAKSHAUG, E., D. A. KIEFER, AND K. ANDRESEN. 1989. A
steady state description of growth and light absorption in
the marine planktonic diatom Skeletonema costatum. Limnol. Oceanogr. 34: 198-205.
SMITH, R.C.,K.S.
BAKER,O.HOLM-HANSEN,AND
R.S.OLSON.
1980. Photoinhibition
of photosynthesis in natural waters.
Photochem. Photobiol. 31: 585-592.
VAN SCOY, K., AND K. COALE. 1994. Pumping iron in the
Pacific. New Sci. 3 Dee, p. 32-35.
WALSH, J. J. 1976. Herbivory as a factor in patterns of nutrient
utilization in the sea. Limnol. Oceanogr. 21: 1-13.
WATSON, A. J., AND OTHERS. 1994. Minimal effect of iron
fertilization on sea-surface carbon dioxide concentrations.
Nature 371: 143-145.
~
P. LISS, AND R. DUCE. 199 1. Design of a small-scale
in $tu iron fertilization experiment. Limnol. Oceanogr. 36:
1960-1965.
WELLS, M. L., N. M. PRICE, AND K. W. BRULAND. 1994. Iron
limitation and the cyanobacterium Synechococcus in equatorial Pacific waters. Limnol. Oceanogr. 39: 1481-1486.